Merge pull request #60 from Addalin/adi

Adi

CameraNetwork/export.py

@@ -1,38 +1,38 @@
-##
-## Copyright (C) 2017, Amit Aides, all rights reserved.
-## 
-## This file is part of Camera Network
-## (see https://bitbucket.org/amitibo/cameranetwork_git).
-## 
-## Redistribution and use in source and binary forms, with or without modification,
-## are permitted provided that the following conditions are met:
-## 
-## 1)  The software is provided under the terms of this license strictly for
-##     academic, non-commercial, not-for-profit purposes.
-## 2)  Redistributions of source code must retain the above copyright notice, this
-##     list of conditions (license) and the following disclaimer.
-## 3)  Redistributions in binary form must reproduce the above copyright notice,
-##     this list of conditions (license) and the following disclaimer in the
-##     documentation and/or other materials provided with the distribution.
-## 4)  The name of the author may not be used to endorse or promote products derived
-##     from this software without specific prior written permission.
-## 5)  As this software depends on other libraries, the user must adhere to and keep
-##     in place any licensing terms of those libraries.
-## 6)  Any publications arising from the use of this software, including but not
-##     limited to academic journal and conference publications, technical reports and
-##     manuals, must cite the following works:
-##     Dmitry Veikherman, Amit Aides, Yoav Y. Schechner and Aviad Levis, "Clouds in The Cloud" Proc. ACCV, pp. 659-674 (2014).
-## 
-## THIS SOFTWARE IS PROVIDED BY THE AUTHOR "AS IS" AND ANY EXPRESS OR IMPLIED
-## WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
-## MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
-## EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT,
-## INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
-## BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-## DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
-## LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
-## OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
-## ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.##
+##
+## Copyright (C) 2017, Amit Aides, all rights reserved.
+## 
+## This file is part of Camera Network
+## (see https://bitbucket.org/amitibo/cameranetwork_git).
+## 
+## Redistribution and use in source and binary forms, with or without modification,
+## are permitted provided that the following conditions are met:
+## 
+## 1)  The software is provided under the terms of this license strictly for
+##     academic, non-commercial, not-for-profit purposes.
+## 2)  Redistributions of source code must retain the above copyright notice, this
+##     list of conditions (license) and the following disclaimer.
+## 3)  Redistributions in binary form must reproduce the above copyright notice,
+##     this list of conditions (license) and the following disclaimer in the
+##     documentation and/or other materials provided with the distribution.
+## 4)  The name of the author may not be used to endorse or promote products derived
+##     from this software without specific prior written permission.
+## 5)  As this software depends on other libraries, the user must adhere to and keep
+##     in place any licensing terms of those libraries.
+## 6)  Any publications arising from the use of this software, including but not
+##     limited to academic journal and conference publications, technical reports and
+##     manuals, must cite the following works:
+##     Dmitry Veikherman, Amit Aides, Yoav Y. Schechner and Aviad Levis, "Clouds in The Cloud" Proc. ACCV, pp. 659-674 (2014).
+## 
+## THIS SOFTWARE IS PROVIDED BY THE AUTHOR "AS IS" AND ANY EXPRESS OR IMPLIED
+## WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+## MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
+## EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT,
+## INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
+## BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+## DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+## LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+## OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+## ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.##
 """Utilites for handling export to solver.
 """
 from __future__ import division
@@ -63,7 +63,7 @@ def exportToShdom(
         base_path (str): Path to store export data.
         array_items (list): List of array items.
         grid (list): List of grid array. This is the grid to reconstruct.
-        lat, lon, lat (float): The latitude, longitude and altitude of the center
+        lat, lon, alt (float): The latitude, longitude and altitude of the center
             of the grid.
         progress_callback (function): Callback function to update the (GUI) with
             the progress of the export.

CameraNetwork/global_settings.py

@@ -264,7 +264,7 @@
 #
 # Amit:
 # The default radiometric settings were taken from camera 109.
-#
+# TODO: what are the units?
 DEFAULT_RADIOMETRIC_SETTINGS = \
     [0.077523644728250449, 0.11163146919636868, 0.11847859109577621]
 

CameraNetwork/gui/main.py

@@ -883,7 +883,7 @@ def projectECEF(self, ECEF_pts, filter_fov=True, errstate='warn'):
 
         Args:
             ECEF_pts (tuple of arrays): points in ECEF coords.
-            fiter_fov (bool, optional): If True, points below the horizion
+            filter_fov (bool, optional): If True, points below the horizion
                will not be returned. If false, the indices of these points
                will be returned.
 
@@ -938,17 +938,20 @@ def _update_cloud_weights(self, change):
         if change["value"] is None:
             return
 
+        # The following calculation is based on page 23 in Amit's thesis (Computation of cloud scores maps)
+        # TODO: consider to change cloud score in 2D .
         r = self.img_array[..., 0].astype(np.float)
         b = self.img_array[..., 2].astype(np.float)
 
         cloud_weights = np.zeros_like(r)
         eps = np.finfo(b.dtype).eps
 
-        threshold = self.cloud_weight_threshold
+        threshold = self.cloud_weight_threshold # currently 0.8
         ratio = r / (b+eps)
         ratio_mask = ratio>threshold
         cloud_weights[ratio_mask] = \
             (2-threshold)/(1-threshold)*(ratio[ratio_mask]-threshold)/(ratio[ratio_mask]+1-threshold)
+        #The current calculation:  ( 1.2 / 0.2 ) * (r - 0.8 ) / (r + 0.2 )
 
         #
         # Limit cloud_weights to 1.
@@ -1403,15 +1406,15 @@ class MainModel(Atom):
     #
     # Reconstruction Grid parameters.
     # Note:
-    # There are two grids used:
+    # There are three grids used:
     # - GRID_VIS_ECEF: Used for visualization on the camera array.
     # - GRID_ECEF: Used for the visual hull algorithm.
     # - GRID_NED: The grid exported for reconstruction.
     #
-    delx = Float(150)
-    dely = Float(150)
-    delz = Float(100)
-    TOG = Float(12000)
+    delx = Float(150) #  [meters]
+    dely = Float(150) #  [meters]
+    delz = Float(100) #  [meters]
+    TOG = Float(12000) #  [meters]
     GRID_VIS_ECEF = Tuple()
     GRID_ECEF = Tuple()
     GRID_NED = Tuple()

CameraNetwork/image_utils.py

@@ -33,7 +33,8 @@
 ## LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
 ## OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
 ## ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.##
-"""General utilities for image processing.
+"""
+General utilities for image processing.
 """
 from __future__ import division, absolute_import, print_function
 from CameraNetwork.utils import obj
@@ -701,23 +702,26 @@ def calc_visual_hull(
         # Calculate the collective clouds weight.
         #
         weights = np.array(grid_scores).prod(axis=0)
-
+        # TODO: Consider to change the prod to other arithmethic calculation. (maybe OR ?) (I think this removes clouds edges from the space carving)
         #
         # voxels that are not seen (outside the fov/sun_mask) by at least two cameras
         # are zeroed.
         #
         grid_masks = np.array(grid_masks).sum(axis=0)
         weights[grid_masks<2] = 0
+
+        #
+        # More cameras viewing the voxel increase it's own score.
+        #
         nzi = weights > 0
         weights[nzi] = weights[nzi]**(1/grid_masks[nzi])
 
         #
-        # Calculate color consistency as described in the article
+        # Calculate color consistency as described in the article. (see Equation 5.5 in p.54 , Amit's thesis)
         #
         std_rgb = np.dstack(cloud_rgb).std(axis=2).mean(axis=1)
         mean_rgb = np.dstack(cloud_rgb).mean(axis=2).mean(axis=1)
-        color_consistency = np.exp(-(std_rgb/mean_rgb)/color_consistency_sigma)
-
+        color_consistency = np.exp(-(std_rgb/(mean_rgb + np.finfo(float).eps))/color_consistency_sigma)
         #
         # Take into account both the clouds weights and photo consistency.
         #

Image_pipeline.md

@@ -138,7 +138,33 @@ For example:
 2. What inputs/changes are required for a new experiment?
 
 ###4. 3D grid and space curving:
+The [geographic coordinate systems](https://en.wikipedia.org/wiki/Geographic_coordinate_system) that are used here are: 
+1. The ECEF  is version used as the common 3D grid that is being used for moving the view-of-point arround the grid conveniently according to cameras positions. 
+2. The NED (X,Y,Z) is used for visualization and reconstruction grid.
 
+See their definitions in the project [here](https://github.com/Addalin/cameranetwork/blob/c69dda2adc041dc2dc98660b34e57769213f23a9/CameraNetwork/gui/main.py#L1393-L1420). 
+
+There are several conversion processes that are being done: 
+
+1. [ProjectGrid()](https://github.com/Addalin/cameranetwork/blob/fa7d2b2f29d5217cdc2b216ae55d147393e9db0d/CameraNetwork/image_utils.py#L615-L645) - Projecting the 3D grid of the interest volume, onto image plane. Which uses ecef2ned in [projectECEF()](https://github.com/Addalin/cameranetwork/blob/c69dda2adc041dc2dc98660b34e57769213f23a9/CameraNetwork/gui/main.py#L881-L933). 
+The 3D NED grid is of size 12 X 12 X 10 [km^3], having 81 X 81 X 121 voxels, each voxel size is of 150 X 150 X 100 [m^3].
+The grid is projected to a 2d grid, shown as red dots on image plane).
+This is done when choosing: `View settings`-->`Widgets`--> `show grid`. 
+// TODO : add snapshots of the 3d view and image with red dotted grid. 
+This method is also being used when computing the [space carve](https://github.com/Addalin/cameranetwork/blob/fa7d2b2f29d5217cdc2b216ae55d147393e9db0d/CameraNetwork/image_utils.py#L738-L810) score per each view.
+This is done when choosing in the map view `Space carving`-->`Show space carving`.
+// TODO : add snapshots of the map with the space carving. 
+Another usage of this method is when applying [Update LOS](https://github.com/Addalin/cameranetwork/blob/c69dda2adc041dc2dc98660b34e57769213f23a9/CameraNetwork/gui/main.py#L620-L667). Ths function converts the  also uses 
+LOS of a single image to the epipolar lines on all other images.
+// TODO : add snapshot of this optsion and how to get to it. 
+
+2. [Space carving](https://github.com/Addalin/cameranetwork/blob/c69dda2adc041dc2dc98660b34e57769213f23a9/CameraNetwork/gui/main.py#L317-L337) - for the grid.  (// TODOL add snapshot and explantaion here.). Calls for space carve on each view separatly using proccess pool. 
+
+[Space carving](https://github.com/Addalin/cameranetwork/blob/fa7d2b2f29d5217cdc2b216ae55d147393e9db0d/CameraNetwork/image_utils.py#L739-L810) - for each view, projecting the the 3d grid onto the image plane. 
+This process is done according to the no. of pertubations chosen by the user (// TODOL add snapshot and explantaion here.). At the end receiving mean of scores for the voxels that are seen from this view (a.k.a `grid_score`). The grid score is based on cloud score that is done on the 2D image. 
+[The cloud scoring](https://github.com/Addalin/cameranetwork/blob/c69dda2adc041dc2dc98660b34e57769213f23a9/CameraNetwork/gui/main.py#L936-L987) in the image plane is based on page 23 in Amit's thesis (Computation of cloud scores maps) ( cloud sore is caled `cloud_weights`)
+
+Then the grid level method, collects scores from all servers. 
 
 ### TODO: Other issues to cover regarding image pipeline: 
 1. Space curving - the transition from 2d and 3d.